Functions and Requirements of Direct-Off-Line SMPS -- COMMON REQUIREMENTS: AN OVERVIEW

1. INTRODUCTION

The "direct-off-line" switchmode supply is so called because it takes its power input directly from the ac power lines, without using the rather large low-frequency (60 to 50 Hz) isolation transformer normally found in linear power supplies.

Although the various switchmode conversion techniques are often very different in terms of circuit design, they have, over many years, developed very similar basic functional characteristics that have become generally accepted industry standards.

Further, the need to satisfy various national and international safety, electromagnetic compatibility, and line transient requirements has forced the adoption of relatively standard techniques for track and component spacing, noise filter design, and transient protection.

The prudent designer will be familiar with all these agency needs before proceeding with a design. Many otherwise sound designs have failed as a result of their inability to satisfy safety agency standards.

Many of the requirements outlined in this section will be common to all switching sup plies, irrespective of the design strategy or circuit. Although the functions tend to remain the same for all units, the circuit techniques used to obtain them may be quite different.

There are many ways of meeting these needs, and there will usually be a best approach for a particular application.

The designer must also consider all the minor facets of the specification before deciding on a design strategy. Failure to consider at an early stage some very minor system requirement could completely negate a design approach-for example, power good and power failure indicators and signals, which require an auxiliary supply irrespective of the converter action, would completely negate a design approach which does not provide this auxiliary supply when the converter is inhibited! It can often prove to be very difficult to provide for some minor neglected need at the end of the design and development exercise.

The remainder of Section 1 gives an overview of the basic input and output functions most often required by the user or specified by national or international standards. They will assist in the checking or development of the initial specification, and all should be considered before moving to the design stage.

2. INPUT TRANSIENT VOLTAGE PROTECTION

Both artificial and naturally occurring electrical phenomena cause very large transient volt ages on all but fully conditioned supply lines from time to time.

IEEE Standard 587-1980 shows the results of an investigation of this phenomenon at various locations. These are classified as low-stress class A, medium-stress class B, and high-stress class C locations. Most power supplies will be in low- and medium-risk locations, where stress levels may reach 6000 V at up to 3000 A.

Power supplies are often required to protect themselves and the end equipment from these stress conditions. To meet this need requires special protection devices.

3. ELECTROMAGNETIC COMPATIBILITY

Input Filters

Switching power supplies are electrically noisy, and to meet the requirements of the various national and international RFI (radio-frequency interference) regulations for conducted mode noise, a differential- and common-mode noise filter is normally fitted in series with the line inputs. The attenuation factor required from this noise filter depends on the power supply size, operating frequency, power supply design, application, and environment.

For domestic and office equipment, such as personal computers, VDUs, and so on, the more stringent regulations apply, and FCC class B or similar limits would normally be applied. For industrial applications, the less severe FCC class A or similar limits would apply.

It is important to appreciate that it is very difficult to cure a badly designed supply by fitting filters. The need for minimum noise coupling must be considered at all stages of the design; some good guidelines are covered in Part 1, Sections 3 and 4.

4. DIFFERENTIAL-MODE NOISE

Differential-mode noise refers to the component of high-frequency electrical noise between any two supply or output lines. For example, this would be measured between the live and neutral input lines or between the positive and negative output lines.

5. COMMON-MODE NOISE

For the line input, common-mode noise refers to that component of electrical noise that exists between both supply lines (in common) and the earth (ground) return.

For the outputs, the position is more complicated, as various configurations of isolated and nonisolated connections are possible. In general, output common-mode noise refers to the electrical noise between any output and some common point, usually the chassis or common return line.

Some specifications, notably those applying to medical electronics, severely limit the amount of ground return current permitted between either supply line and the earth (ground) return. A ground return current normally flows through the filter capacitors and leakage capacitance to ground, even if the insulation is perfect. The return current limitation can have a significant effect on the design of the supply and the size of input filter capacitors. In any event, capacitors in excess of 0.01 MF between the live line and ground are not permitted by many safety standards.

6. FARADAY SCREENS

High-frequency conducted-mode noise (noise conducted along the supply or output leads) is normally caused by capacitively coupled currents in the ground plane or between input and output circuits. For this reason, high-voltage switching devices should not be mounted on the chassis. Where this cannot be avoided, a Faraday screen should be fitted between the noise source and the ground plane, or at least the capacitance to the chassis should be minimized.

To reduce input-to-output noise coupling in isolating transformers, Faraday screens should be fitted. These should not be confused with the more familiar safety screens.

7. INPUT FUSE SELECTION

The fuse is an often neglected part of power supply design. Modern fuse technology makes available a wide range of fuses designed to satisfy closely defined parameters. Voltages, inrush currents, continuous currents, and let-through energy (I^2 t ratings) should all be considered. Where units are dual-input-voltage-rated, it may be necessary to use a lower fuse current rating for the higher input voltage condition. Standard, medium-speed glass cartridge fuses are universally available and are best used where possible. For line input applications, the current rating should take into account the 0.6 to 0.7 power factor of the capacitive input filter used in most switchmode systems.

For best protection the input fuse should have the minimum current rating that will reliably sustain the inrush current and maximum operating currents of the supply at minimum line inputs. However, it should be noted that the rated fuse current given in the fuse manufacturer's data is for a limited service life, typically a thousand hours operation. For long fuse life, the normal power supply current should be well below the maximum fuse rating; the larger the margin, the longer the fuse life.

Fuse selection is therefore a compromise between long life and full protection. Users should be aware that fuses wear with age and should be replaced at routine servicing periods. For maximum safety during fuse replacement, the live input is normally fused at a point after the input switch.

To satisfy safety agency requirements and maintain maximum protection, when fuses are replaced, a fuse of the same type and rating must be used.

8. LINE RECTIFICATION AND CAPACITOR INPUT FILTERS

Rectifier capacitor input filters have become almost universal for direct-off-line switch mode power supplies. In such systems the line input is directly rectified into a large electrolytic reservoir capacitor.

Although this circuit is small, efficient, and low-cost, it has the disadvantage of demanding short, high-current pulses at the peak of the applied sine-wave input, causing excessive line I^2 R losses, harmonic distortion, and a low power factor.

In some applications (e.g., shipboard equipment), this current distortion cannot be tolerated, and special low-distortion input circuits must be used.

9. INRUSH LIMITING

Inrush limiting reduces the current flowing into the input terminals when the supply is first switched on. It should not be confused with "soft start," which is a separate function controlling the way the power converter starts its switching action.

In the interests of minimum size and weight, most switchmode supplies use semiconductor rectifiers and low-impedance input electrolytics in a capacitive input filter configuration. Such systems have an inherently low input resistance; also, because the capacitors are initially discharged, very large surge currents would occur at switch-on if such filters were switched directly to the line input.

Hence, it is normal practice to provide some form of current inrush limiting on power supplies that have capacitive input filters. This inrush limiting typically takes the form of a resistive limiting device in series with the supply lines. In high-power systems, the limiting resistance would normally be removed (shorted out) by an SCR, triac, or switch when the input reservoir and/or filter capacitor has been fully charged. In low-power systems, NTC thermistors are often used as limiting devices.

The selection of the inrush-limiting resistance value is usually a compromise between acceptable inrush current amplitude and start-up delay time. Negative temperature coefficient thermistors are often used in low-power applications, but it should be noted that thermistors will not always give full inrush limiting. For example, if, after the power supply has been running long enough for the thermistor to heat up, the input is turned rapidly off and back on again, the thermistor will still be hot and hence low-resistance, and the inrush current will be large. The published specification should reflect this effect, as it is up to the user to decide whether this limitation will cause any operational problems. Since even with a hot NTC the inrush current will not normally be damaging to the supply, thermistors are usually acceptable and are often used for low-power applications.

10. START-UP METHODS

In direct-off-line switchmode supplies, the elimination of the low-frequency (50 to 60 Hz) transformer can present problems with system start-up. The difficulty usually stems from the fact that the high-frequency power transformer cannot be used for auxiliary supplies until the converter has started. Suitable start-up circuits are discussed in Part 1, Section 8.

11. SOFT START

Soft start is the term used to describe a low-stress start-up action, normally applied to the pulse-width-modulated converter to reduce transformer and output capacitor stress and to reduce the surge on the input circuits when the converter action starts.

Ideally, the input reservoir capacitors should be fully charged before converter action commences; hence, the converter start-up should be delayed for several line cycles, then start with a narrow, progressively increasing pulse width until the output is established.

There are a number of reasons why the pulse width should be narrow when the converter starts, and progressively increase during the start-up phase. There will often be consider able capacitance on the output lines, and this should be charged slowly so that it does not reflect an excessive transient back to the supply lines. Further, where a push-pull action is applied to the main transformer, flux doubling and possible saturation of the core may occur if a wide pulse is applied to the transformer for the first half cycle of operation. Finally, since an inductor will invariably appear somewhere in series with the cur rent path, it may be impossible to prevent voltage overshoot on the output if this inductor current is allowed to rise to a high value during the start-up phase.

12. START-UP OVERVOLTAGE PREVENTION

When the power supply is first switched on, the control and regulator circuits are not in their normal working condition (unless they were previously energized by some auxiliary supply).

As a result of the limited output range of the control and driver circuits, the large-signal slew rate may be very nonlinear and slow. Hence, during the start-up phase, a "race" condition can exist between the establishment of the output voltages and correct operation of the control circuits. This can result in excessive output voltage overshoot.

Additional fast-acting voltage clamping circuits may be required to prevent overshoot during the start-up phase, a need often overlooked in the past by designers of both discrete and integrated control circuits.

13. OUTPUT OVERVOLTAGE PROTECTION

Loss of voltage control can result in excessive output voltages in both linear and switch mode supplies. In the linear supply (and some switching regulators), there is a direct DC link between input and output circuits, so that a short circuit of the power control device results in a large and uncontrolled output. Such circuits require a powerful overvoltage clamping technique, and typically an SCR "crowbar" will short-circuit the output and open a series fuse.

In the direct-off-line SMPS, the output is isolated from the input by a well-insulated transformer. In such systems, most failures result in a low or zero output voltage. The need for crowbar-type protection is less marked, and indeed is often considered incompatible with size limitations. In such systems, an independent signal level voltage clamp which acts on the converter drive circuit is often considered satisfactory for overvoltage protection.

The design aim is that a single component failure within the supply will not cause an overvoltage condition. Since this aim is rarely fully satisfied by the signal level clamping techniques often used (for example, an insulation failure is not fully protected), the crowbar and fuse technique should still be considered for the most exacting switchmode designs. The crowbar also provides some protection against externally induced overvoltage conditions.

14. OUTPUT UNDERVOLTAGE PROTECTION

Output undervoltages can be caused by excessive transient current demands and power out ages. In switchmode supplies, considerable energy is often stored in the input capacitors, and this provides "holdup" of the outputs during short power outages. However, transient current demands can still cause under-voltages as a result of limited current ratings and output line voltage drop. In systems that are subject to large transient demands, the active undervoltage prevention circuit described in Part 1, Section 12 should be considered.

15. OVERLOAD PROTECTION (INPUT POWER LIMITING)

Power limiting is usually applied to the primary circuits and is concerned with limiting the maximum throughput power of the power converter. In multiple-output converters this is often necessary because, in the interest of maximum versatility, the sum of the independent output current limits often has a total VA rating in excess of the maximum converter capability.

Primary power limiting is often provided as additional backup protection, even where normal output current limiting would prevent output overloading conditions. Fast-acting primary limiting has the advantage of preventing power device failure under unusual transient loading conditions, when the normal secondary current limiting may not be fast enough to be fully effective. Furthermore, the risk of fire or excessive power supply damage in the event of a component failure is reduced. Power supplies with primary power limiting usually have a much higher reliability record than those without this additional protection.

16. OUTPUT CURRENT LIMITING

In higher-power switchmode units, each output line is independently current-limited. The current limit should protect the supply under all load conditions including short-circuit.

Continuous operation in a current-limited mode should not cause overdissipation or failure of the power supply. The switchmode unit (unlike the linear regulator) should have a constant current limit. By its nature, the switching supply does not dissipate excessive power under short-circuit conditions, and a constant current limit is far less likely to give the user such problems as "lockout" under nonlinear or cross-coupled load conditions. (Cross-coupled loads are loads that are connected between a positive and a negative output line without connection to the common line.) Linear regulators traditionally have reentrant (foldback) current limiting in order to prevent excessive dissipation in the series element under short-circuit conditions. Section 14.5 covers the problems associated with cross-coupled loads and reentrant current limits more fully.

17. BASE DRIVE REQUIREMENTS FOR HIGH-VOLTAGE BIPOLAR TRANSISTORS

In direct-off-line SMPSs the voltage stress on the main switching devices can be very large, of the order of 800 to 1000 V in the case of the flyback converter.

Apart from the obvious needs for high-voltage transistors, "snubber" networks, load line shaping, and antisaturation diodes, many devices require base drive waveform shaping.

In particular, the base current is often required to ramp down during the turn-off edge at a controlled rate for best performance.

18. PROPORTIONAL DRIVE CIRCUITS

With bipolar transistors, base drive currents in excess of those required to saturate the transistor reduce efficiency and can cause excessive turn-off storage times with reduced control at light loads.

Improved performance can be obtained by making the base drive current proportional to the collector current. Suitable circuits are shown in Part 1, Section 16.

19. ANTISATURATION TECHNIQUES

With bipolar transistors, in the switching mode, improved turn-off performance can be obtained by preventing "hard" saturation. The transistor can be maintained in a quasi saturated state by maintaining the drive current at a minimum defined by the gain and collector current. However, since the gain of the transistor changes with device, load, and temperature, a dynamic control is required.

Antisaturation circuits are often combined with proportional drive techniques. Suitable methods are shown in Part 1, Section 17.

20. SNUBBER NETWORKS

This is a power supply engineering term used to describe networks which provide turn-on and turn-off load line shaping for a switching device.

Load line shaping is required to prevent breakdown by maintaining the switching device within its "safe operating area" throughout the switching cycle.

In many cases snubber networks also reduce RFI problems as a result of the reduced dv/dt on switching elements, although this is not their primary function.

21. CROSS CONDUCTION

In half-bridge, full-bridge, and push-pull applications, a DC path exists between the supply lines if the "on" states of the two switching devices overlap. This is called "cross conduction" and can cause immediate failure.

To prevent this condition, a "dead time" (a period when both devices are off) is often provided in the drive waveform. To maintain full-range pulse-width control, a dynamic dead time may be provided.

22. OUTPUT FILTERING, COMMON-MODE NOISE, AND INPUT-TO-OUTPUT ISOLATION

These parameters have been linked together, as they tend to be mutually interdependent. In switchmode supplies, high voltages and high currents are being switched at very fast rates of change at ever-increasing frequencies. This gives rise to electrostatic and electromagnetic radiation within the power supply. The electrostatic coupling between high-voltage switching elements and the output circuit or ground can produce particularly troublesome common-mode noise problems.

The problems associated with common-mode noise are sometimes not recognized, and there is a tendency to leave this requirement out of the power supply specifications.

Common-mode noise can cause system problems, and it is good power supply design practice to minimize the capacitance between the switching elements and chassis and to provide Faraday screens between the primary and secondary of the power transformer.

Where switching elements are to be mounted on the chassis for cooling purposes, an insulated Faraday screen should be placed between the switching element and the mounting surface. This screen and any other Faraday screens in the transformer should be returned to one of the input DC supply lines so as to return capacitively coupled currents to the source.

In many cases, the transformer will require an additional safety screen connected to earth or chassis. This safety screen should be positioned between the RF Faraday screen and the output windings.

In rare cases (where the output voltages are high), a second Faraday screen may be required between the safety screen and the output windings to reduce output common mode current. This screen should be returned to the common output line, as close as possible to the transformer common-line connection pin.

The screens, together with the necessary insulation, increase the spacing between the primary and secondary windings, thereby increasing the leakage inductance and degrading transformer performance. It should be noted that the Faraday screen does not need to meet the high current capacity of the safety screens and therefore can be made from lightweight material and connections.

23. POWER FAILURE SIGNALS

To allow time for "housekeeping" functions in computer systems, a warning of impending shutdown is often required from the power supply. Various methods are used, and typically a warning signal should be given at least 5 ms before the power supply outputs fall below their minimum specified values. This is required to allow time for a controlled shutdown of the computer.

In many cases, simple power failure systems are used that recognize the presence or absence of the AC line input and give a TTL low signal within a few milliseconds of line failure. It should be recognized that the line input passes through zero twice in each cycle under normal conditions; since this must not be recognized as a failure, there is usually a delay of several milliseconds before a genuine failure can be recognized. When a line failure has been recognized, the normal holdup time of the power supply should provide output voltage for a further period, allowing time for the necessary housekeeping procedures.

Two undesirable limitations of these simple systems should be recognized. First, if a "brownout" condition precedes the power failure, the output voltage may fall below the minimum value without a power failure signal being generated. Second, if the line input voltage to the power supply immediately prior to failure is close to the minimum required for normal operation, the holdup time will be severely diminished, and the time between a power failure warning and supply shutdown may not be long enough for effective housekeeping.

For critical applications, more sophisticated power failure warning systems that recognize brownout should be used. If additional holdup time is required, charge dumping techniques should be considered.

24. POWER GOOD SIGNALS

"Power good" signals are sometimes required from the power supply. These are usually TTL-compatible outputs that go to a "power good" (high state) when all power supply volt ages are within their specified operating window. "Power good" and "power failure" signals are sometimes combined. LED (light-emitting diode) status indicators are often provided with the "power good" signals, to give a visual indication of the power supply status.

25. DUAL INPUT VOLTAGE OPERATION

With the trend toward international trading it is becoming increasingly necessary to provide switchmode supplies for dual input, nominally 110/220-V operation. A wide variety of techniques are used to meet these dual-voltage requirements, including single or multiple transformer tap changes that may be manual or automatic, and selectable voltage doubling.

If auxiliary transformers and cooling fans are used, these must be considered in the dual voltage connection.

A useful method of avoiding the need for special dual-voltage fans and auxiliary transformers is shown in Part 1, Section 23. It should be remembered that the insulation of the auxiliary transformer and fan must meet the safety requirements for the highest-voltage input. More recently, high-efficiency "brushless" DC fans have become available; these can be driven by the supply output, overcoming insulation and tap change problems.

The voltage doubler technique with one or two link changes is probably the most cost effective and is generally favored in switchmode supplies. However, when this method is used, the design of the filter, the input fuse, and inrush limiting should be considered.

When changing the input voltage links the low-voltage tap position gives higher current stress, whereas the higher tap position gives greater voltage stress. The need to meet both conditions results in more expensive filter components. Therefore, dual-voltage operation should not be specified unless this is a real system requirement.

26. POWER SUPPLY HOLDUP TIME

One of the major advantages of switchmode over linear supplies is their ability to maintain the output voltages constant for a short period after line failure. This "holdup time" is typically 20 ms, but depends on when during the input cycle the power failure occurs and the loading and the supply voltage before the line failure.

A major factor controlling the holdup time is the history and amplitude of the supply voltage immediately prior to the failure condition. Most specifications define holdup time from nominal input voltage and loading. Holdup times may be considerably less if the sup ply voltage is close to its minimum value immediately prior to failure.

Power supplies that are specified for long holdup times at minimum input voltages are either expensive because of the increased size of input capacitors, or less efficient because the power converter must now maintain the output voltage constant for a much lower input voltage. This usually results in less efficient operation at nominal line inputs. Charge dumping techniques should be considered when long holdup times are required at low input voltages.

27. SYNCHRONIZATION

Synchronization of the switching frequency is sometimes called for, particularly when the supply is to be used for VDU (visual display unit) applications. Although synchronization is of dubious value in most cases, as adequate screening and filtering of the supply should eliminate the need, it must be recognized that systems engineers often specify it.

The constraints placed on the power supply design by specifying synchronization are severe; for example, low-cost variable frequency systems cannot be used. Furthermore, the synchronization port gives access to the drive circuit of the main converter and provides a means whereby the operating integrity of the converter can be disrupted.

The possibility of badly defined or incorrect synchronization information must be considered in the design of synchronizable systems; techniques used should be as insensitive as possible to abuse. The system engineer should be aware that it is difficult to guarantee that a power supply will not be damaged by incorrect or badly defined synchronization signals. Because of the need to prevent saturation in wound components, most switchmode supplies use oscillator designs which can be synchronized only to frequencies higher than the natural oscillation frequency. Also, the synchronization range is often quite limited.

28. EXTERNAL INHIBIT

For system control, it is often necessary to turn the power supply on or off by external electronic means. Typically a TTL high signal will define the "on" condition and a TTL low the "off' condition. Activation of this electronic inhibit should invoke the normal soft start sequence of the power supply when it is turned on. Power supplies for which this remote control function is required often need internal auxiliary supplies that are common to the output. The auxiliary supply must be present irrespective of main converter operation. This apparently simple requirement may define the complete design strategy for the auxiliary supplies.

29. FORCED CURRENT SHARING

Voltage-controlled power supplies, by their nature, are low-output-impedance devices.

Since the output voltage and performance characteristics of two or more units will never be identical, the units will not naturally share the load current when they are operated in parallel.

Various methods are used to force current sharing. However, in most cases these techniques force current sharing by degrading the output impedance (and consequently the load regulation) of the supply. Hence the load regulation performance in parallel forced current sharing applications will usually be lower than that found with a single unit.

A possible exception is the master-slave technique which tends to a voltage-controlled current source. However, the master-slave technique has the disadvantage of its inability to provide good parallel redundant operation. A failure of the master system usually results in complete system failure.

Interconnected systems of current-mode control topologies can resolve some of these problems, but the tendency for noise pickup on the P-terminal (parallel current sharing) link between units makes it somewhat difficult to implement in practice. Further, if one unit is used to provide the control signal, failure of this unit will shut down the whole system, which is again contrary to the needs of a parallel redundant system.

The forced current sharing system described in Part 1, Section 24 does not suffer from these difficulties. Although the output regulation is slightly degraded, the variation in output voltage in normal circumstances is only a few millivolts, which should be acceptable for most practical applications.

Failure to provide current sharing means that one or more of the power supplies will be operating in a maximum current limited mode, while others are hardly loaded. However, so long as the current limits for the units are set at a value where continuous operation in the current limited mode gives reasonable power supply life, simple direct parallel connection can be used.

30. REMOTE SENSING

If the load is situated some distance from the power supply, and the supply-lead voltage drops are significant, improved performance will be obtained if a remote voltage sense is used for the power supply. In principle, the reference voltage and amplifier comparator inputs are connected to the remote load by voltage sensing lines separate from the current carrying lines to remove the line-drop effects. These remote sense leads carry negligible currents, so the voltage drop is also negligible. This arrangement permits the power supply to compensate for the voltage drops in the output power leads by increasing the supply volt age as required. In low-voltage, high-current applications, this facility is particularly useful.

However, the user should be aware of at least three limitations of this technique:

1. The maximum external voltage drop that can be tolerated in the supply leads is typically limited to 250 mV in both go and return leads (500 mV overall). In a 100-A 5-V application, this represents an extra 50 W from the power supply, and it should be remembered that this power is being dissipated in the supply lines.

2. Where power supplies are to be connected in a parallel redundant mode, it is common practice to isolate each supply with a series diode. The principle here is that if one power supply should short-circuit, the diode will isolate this supply from the remaining units.

If this connection is used, then the voltage at the terminals of the power supply must be at least 0.7 V higher than the load, neglecting any lead losses, and the required terminal voltage may exceed the power supply's design maximum unless the supply is specifically designed for this mode of operation. Furthermore, it must be borne in mind that in the event of power supply failure in this parallel redundant mode, the amplifier sense leads will still be connected to the load and will experience the load voltage. The remote sense circuit must be able to sustain this condition without further damage.

It is common practice to link the remote sense terminals to the power supply output terminals with resistors within the supply to prevent loss of control and voltage over shoot in the event of the sense leads being disconnected. Where such resistors are used in the parallel redundant connection, they must be able to dissipate the appropriate power, VR out /2, without failure in the event of the main terminal output voltage falling to zero.

3. Remote sense terminals are connected to a high-gain part of the power amplifier loop.

Consequently, any noise picked up in the remote sense leads will be translated as output voltage noise to the power supply terminals, degrading the performance. Further, the additional phase shifts caused by lead inductance and resistance can have a destabilizing effect. Therefore, it is recommended that remote sense leads be twisted to minimize inductance and noise pickup.

Unless they are correctly matched and terminated, coaxial leads are not recommended, as the distributed capacitance can degrade the transient performance.

31. P-TERMINAL LINK

In power supplies where provision is made to interlink one or more units in a parallel forced current sharing mode, current sharing communication between supplies is required.

This link is normally referred to as the P-terminal link. In master-slave applications this link allows the master to control the output regulators of the slave units. In forced current sharing applications this link provides communication between the power supplies, indicating the average load current and allowing each supply to adjust its output to the correct proportion of the total load. Once again, the P-terminal link is a noise-sensitive input, so the connections should be routed so as to minimize the noise pickup.

32. LOW-VOLTAGE CUTOUT

In most applications, the auxiliary supplies to the power unit are derived from the same sup ply lines as the main converter. For the converter to start up under controlled conditions, it is necessary that the supply to both the main converter and the auxiliary circuits be correctly conditioned before the power converter action commences. It is normal practice to provide a drive inhibiting circuit which is activated when the auxiliary supply falls below a value which can guarantee proper operation. This "low-voltage inhibit" prevents the converter from starting up during the power-up phase until the supply voltage is sufficiently high to ensure proper operation. Once the converter is running, if the supply voltage falls below a second, lower value, the converter action will be inhibited; this hysteresis is provided to prevent squegging at the threshold voltage.

33. VOLTAGE AND CURRENT LIMIT ADJUSTMENTS

The use of potentiometers for voltage and current limit adjustments is not recommended, except for initial prototype applications. Power supply voltages and current limits, once set, are very rarely adjusted. Most potentiometers become noisy and unreliable unless they are periodically exercised, and this causes noisy and unreliable performance. Where adjustments are to be provided, high-grade potentiometers must be used.

34. INPUT SAFETY REQUIREMENTS

Most countries have strict regulations governing safety in electrical apparatus, including power supplies. UL (Underwriters Laboratories), VDE (Verband Deutscher Elektrotechniker), IEC (International Electrotechnical Commission), and CSA (Canadian Standards Association) are typical examples of the bodies formulating these regulations.

It should be remembered that these regulations define minimum insulation, spacing, and creepage distance requirements for printed circuit boards, transformers, and other components.

Meeting these specifications will have an impact on performance and must be an integral part of the design exercise. It is very difficult to modify units to meet safety regulations after they have been designed. Consequently, drawing office and design staff should be continually alert to these requirements during the design phase. Furthermore, the technical requirements for high performance tend to be incompatible with the spacing requirements for the safety specifications. Consequently, a prototype unit designed without full attention to the safety spacing needs may give an excessively optimistic view of performance which cannot be maintained in the fully approvable finished product.

A requirement often neglected is that ground wires, safety screens, and screen connections must be capable of carrying the fuse fault current without rupture, to prevent loss of safety ground connections under fault conditions. Further, any removable mountings (which, for example, may have been used to provide an earth connection from the printed circuit board to earth or chassis) must have a provision for hard wiring of the ground of the host equipment main frame. Mounting screws alone do not meet the safety requirements for some authorities.

Also see: Our other Switching Power Supply Guide